Start a Conversation

This post is more than 5 years old

Solved!

Go to Solution

4246

January 13th, 2012 17:00

Disk Array Layout

Hello,

Our current setup is as follows:

VNX5100

1 x DPE (15 sas drives - 4 for OS)

1 x DAE (15 sas drives)

1 x DAE (2 sas drives)

We are using the array for a VMware server farm and was wondering what a recommended configuration could be to minimize downtime in case of a lost DAE:

For example moving the disks:

dpe - 10 disks

dae - 9 disks

dae - 9 disks

Then create some RAID groups that use disks from each tray????? I know by default the RAID groups take the disks from the same tray.

Thanks

1 Rookie

 • 

20.4K Posts

January 15th, 2012 18:00

if you look in VNX best practices document, there is actually a section for vertical provisioning. Other then some corner cases, it's not something that you MUST do and if you use virtual pools it does not matter anymore.

1 Rookie

 • 

20.4K Posts

January 13th, 2012 19:00

1 Rookie

 • 

20.4K Posts

January 15th, 2012 17:00

i started with clariion back in the days (fc4700) and have never lost a DAE.

199 Posts

January 15th, 2012 17:00

I strongly recommend to create Raid Group cross as many DAEs as possible. I have seen so many times the whole DAE drives dropped not ready due to LCC/Power/Loop issue. It's not a smart way to put all your eggs in same basket.

31 Posts

January 15th, 2012 18:00

And the best practices guides mentioned above doesn't really say horizontal provisioning (default) of disks is NOT recommended.

199 Posts

January 15th, 2012 19:00

When I was in Symmetrix support team, I saw the whole DAE failed every day for variety of reasons. Sometimes the whole bus failed which includes 2 or 3 DAEs. Because Symmetrix always keep another datacopy on other bus, customer won't experience DA. I am always wondering why CLARiiON would select drives in the same DAE to form Raid Group by default. You can follow EMC BP. But if I have a CLARiiON, I won't create Raid Group on that way.

1 Rookie

 • 

20.4K Posts

January 15th, 2012 20:00

each DAE is connected to different buses on each SP, just like DAE is connected to different DA on the Symm ..so how is that different ?

1 Rookie

 • 

20.4K Posts

January 15th, 2012 21:00

jingyi wrote:

Keep in mind pool is also comprised of multiple Raid Groups. If you lose 2 disks in the same raid group, you will face Data Unavailable.

and if you put your raid group members in different DAEs and you have mutlipe drive failures in your RG you won't exprience DU ?

1 Rookie

 • 

20.4K Posts

January 15th, 2012 21:00

jingyi wrote:

Keep in mind pool is also comprised of multiple Raid Groups.

well aware of private RG, do you think they are spread over multiple DAEs. No

1 Rookie

 • 

20.4K Posts

January 15th, 2012 21:00

DAE on Clariion is connected to one bus on SPA and one bus on SPB,  so even iif you have two disk in the same DAE, they are available to both buses. 

199 Posts

January 15th, 2012 21:00

Keep in mind pool is also comprised of multiple Raid Groups. If you lose 2 disks in the same raid group, you will face Data Unavailable. Not spreading Raid Groups to multiple DAE will have higher chance to encounter dual drive failure.

199 Posts

January 15th, 2012 21:00

The difference is if you create a Raid 1 device, symmetrix will never choose 2 disks in  same DA (Bus on CLARiiON).

199 Posts

January 15th, 2012 21:00

dynamox wrote:

jingyi wrote:

Keep in mind pool is also comprised of multiple Raid Groups. If you lose 2 disks in the same raid group, you will face Data Unavailable.

and if you put your raid group members in different DAEs and you have mutlipe drive failures in your RG you won't exprience DU ?

What I'm saying is if you spread disks of one raid group to multiple buses, you won't experience DU if you lose the whole DAE. Other than that, I agree with you.

1 Rookie

 • 

20.4K Posts

January 15th, 2012 22:00

ok, we are in agreement   then. I can see losing one LCC but two at the same ..must be some really really bad luck.

1.3K Posts

March 3rd, 2012 10:00

dynamox wrote:

DAE on Clariion is connected to one bus on SPA and one bus on SPB,  so even iif you have two disk in the same DAE, they are available to both buses.

isn't a given DAE talk/attach to only a single bus? if the DAE itself is available to multiple buses where is the need to distribute buses comes from ! i did not follow the significance of brining up "SP" here..

No Events found!

Top